Isolation and embryonic expression of an abdominal-A-like gene from the lepidopteran, Manduca sexta

Modifications and specializations of segment character account for most of the enormous taxonomic diversity found in the insects. Underlying this diversity, however, is a striking similarity in the basic metameric body plan. Homeotic genes are involved in establishing segment identity (Morata and Lawrence, 1977) and are likely to play a crucial role in generating the diversity of segments, but what maintains the basic metameric plan is largely unknown.

Drosophila homeotic genes are clustered in the Antennapedia and bithorax gene complexes (Lewis, 1978; Mahaffey and Kaufman, 1987). The initial isolation and sequencing of these genes uncovered a region of sequence homology, the homeobox, found in nearly all the homeotic genes (McGinnis et al. 1984; Scott and Weiner, 1984). This homeobox region is able to bind to DNA, and these proteins act as transcriptional regulators (see reviews by Levine and Hoey, 1988; Hayashi and Scott, 1990). The Drosophila homeotic genes are first expressed at the cellular blastoderm stage as a result of the earlier action of the gap, pair-rule, and polarity genes (Akam, 1987; Ingham, 1988). At this time, each body segment is characterized by the expression of an unique set of homeotic genes, the products of which then are thought to control the subsequent expression of segment identity (Garcia-Bellido, 1977).

The common occurrence of homeotic mutations in other insects (for a review of early literature, see Ouweneel, 1976; Sibatani, 1980; Beeman, 1987; Booker and Truman, 1989) suggests that the fundamental aspects in the segmentation process that involve homeotic gene function will be conserved throughout the insects. In fact, clusters of homeotic alleles have been genetically identified in Bombyx (Tazima, 1964) and Tribohum (Beeman, 1987). The isolation and expression patterns of homeotic genes from other insects [Apis (Fleig et al. 1988; Walldorf et al. 1989), Schistocerca (Akam et al. 1988; Tear et al. 1990), and Tribolium (Brown et al. 1990)] as well as from other invertebrates and vertebrates (Gehring, 1987; Scott et al. 1989; Kessel and Gruss, 1990) indicates that at least some of the molecular components used to form and maintain repeated elements along a body axis are conserved throughout the eukaryotes.

Experimental and descriptive analyses of insect embryogenesis, however, suggest a divergence in mechanisms of segmentation within the insects. In particular, not all insects form their segments in the same way. Insects can be generally classified into three categories based on the germ anlage: short, intermediate and long (see Sander, 1976). Short and intermediate germ insects are characterized by a progressive determination of the posterior segments, while long germ insects specify the identity of all the segments nearly simultaneously in the cellular blastoderm. In addition, the different types of germ band respond differently to experimental perturbations such as ligation, UV- and X-irradiation, and heat shock (see Kuhn, 1971; Seidel, 1971; and Sander, 1976 for summaries).

Even within these three categories, there is significant diversity of developmental pattern. For instance, the long germ anlage embryo of Drosophila spends approximately 4% of its developmental time in the cellular blastoderm stage (Campos- ortega and Harten-stein, 1985), whereas that of the honeybee Apis mellifera spends 36 % of its developmental time in this stage and does not subsequently undergo germ band elongation (Fleig and Sander, 1986). These observations imply that, in spite of conserved components in the process of segmentation, the specific manner by which a particular insect constructs its overall body plan differs from group to group. These observed differences may result from simple differences in timing of events relative to one another, or they may in fact represent qualitatively different ways of making segments (Meinhardt, 1982, 1986; Sander, 1983, 1984).

The position of the lepidopterans within this system of classification remains uncertain (Sander, 1984). Studies using localized UV irradiation on both a primitive lepidopteran Tineola (Liischer, 1944) and the more derived Bombyx (Myohara and Kiguchi, 1990) suggest that segment determination may occur before or during cellular blastoderm as expected for a long germ insect. Yet recent studies with a fascicuhn-like antibody (Carr and Taghert, 1989) suggest that in another derived lepidopteran, Manduca sexta, there may be a progressive determination of segments as typical of intermediate and short germ insects.

To begin to understand novel mechanisms of segmentation and early lepidopteran embryogenesis, we have utilized sequence homology to isolate segmentation/homeotic genes from Manduca sexta. We report here the isolation of an homeobox-containing gene with significant sequence similarity to the Drosophila homeotic gene abdominal-A (Karch et al. 1990), and show its temporal and spatial pattern of expression during Manduca development.

A Manduca sexta genomic library (Rebers et al 1987) was screened with a 570 bp EcoRl-BamHl fragment isolated from a clone (p904) of the Antennapedw. gene of Drosophila (Garber et al 1983). The probe contained the 180bp homeobox sequence plus 390 bp 3 ′ to the homeobox. The library was screened under low stringency conditions exactly as described by McGinnis et al. (1984). Low stringency washes were carried out with 2 ×SSC (1 ×SSC: 0.15M NaCl, 0 015M sodium citrate), 0.1% SDS, for 4x15 min at room temperature, then 2 ×30 min at 55 °C.

DNA was isolated by the method of McGinnis et al. (1983) from adult Drosophila and fifth instar larval epidermis of Manduca. Genomic DNA was electrophoresed on 10% agarose gels and transferred to Hybond N (Amersham). A 1.2 kb homeobox-containing EcoKl fragment was labeled with pPJdATP (New England Nuclear) by the method of random priming (Feinberg and Vogelstein, 1984). High-stringency hybridizations were earned out with the same buffer as used for the low-stringency hybridization (McGinnis et al 1984), but the hybridization temperature was raised to 42 °C. Filters were washed 4 ×15min at room temperature in 2 ×SSC, 0.1%SDS, followed by two 30mm washes at 65 °C with O.l °SSC, 0.1% SDS

Manduca adults were housed with tobacco plants under a 17L/7D photopenod at 21 °C Egg collections were made from the plants beginning an hour before lights off (Sasaki and Riddiford, 1984) Eggs were stored at 26 °C for the appropriate development time, washed with 50% bleach (concentrated bleach: 5.25% sodium hypochlonte), rinsed with water, then frozen in liquid nitrogen

RNA from embryos and larval tissues was extracted with guanidine isothiocyanate as described by Davis et al (1986). RNA was quantified by UV spectrophotometry, and the quality of the total RNA was detected by visualization with ethidium bromide in a 1% agarose gel Poly(A)+ RNA was selected by ohgo(dT) (Collaborative Research) chromatography (Aviv and Leder, 1972). RNA was separated on 0.66 M formaldehyde gels (Davis et al. 1986), which included 0.66 ng ml”¹ ethidium bromide. The gels were photographed with a red filter under UV light, then transferred to Hybond N (Amersham). Blots were hybridized at 42 °C, in 50% deionized formamide, 5xSSC, 50mM NaPO₄, pH6 5, 250/igmP¹ sonicated, denatured herring sperm DNA, 250/igml”¹ yeast tRNA, lxDenhardt’s solution (Denhardt, 1966). The probe was a 216 bp EcoRl-XhoI fragment, which contained 133 bp of the homeobox and 92 bp 3 ′ to the homeobox. Both the labeling of this fragment and the washing conditions were the same as for the high-stringency Southerns

The Xhol fragments indicated in Fig. 1 were subcloned into M13 vectors, MP18 or MP19 (Yanish-Perron et al. 1985), and sequenced using single-stranded templates by the dideoxy chain termination technique (Sanger et al. 1977), modified for the use of ³²S-dATP (Biggin et al 1983). Overlapping deletions for some of the clones were created by digesting M13 templates with T4 DNA polymerase (Dale et al 1985).

Analysis of the DNA sequence was performed using IBI/Pustell sequence analysis programs (International Biotechnologies, Inc.; Pustell and Kafatos, 1982)

To generate RNA probes to both coding and non-coding strands, the 216 bp EcoRl-Xhol fragment was subcloned into Bluescnpt (Stratagene). This clone was linearized with either EcoRI or Xhol and labeled with ³⁵S-UTP (Amersham, 1300/iCimmor¹). RNA probes with specific activity greater than l ×l^ctsmin”¹^” were made using T7 and T3 polymerases.

Staged Manduca embryos were rinsed briefly in bleach, then rinsed in water, and fixed in 4 % formalin in phosphate-buffered saline (PBS: 1 3 M NaCl, 0 007M Na₂HPO₄, 0 03M NaH₂PO₄) for 1 h. At this point, the embryos were manually dechononated and fixed for an additional 4 –6 h. Dehydration, paraffin embedding, prehybndization and hybridizations were performed according to Angerer et al (1987). Sections were 6 am thick 30 μl of 1 × 10⁵ disints min ⁻¹μl ⁻¹ of probe were applied to each slide. Exposure times vaned between 5 and 14 days. Sections were stained in 0.02% toluidine blue, 0 02 M borate pH9.4 after developing.

Using a Drosophila Antennapedia probe (Garber el al. 1983) containing the 180 bp Antennapedia homeobox and 390 bp 3 ′ to the homeobox, we screened three genomic equivalents of a Manduca genomic library under low-stringency conditions. We isolated six clones; restriction enzyme mapping showed all of these clones to be overlapping or duplicates, and to encompass a 15 kb genomic region. The region of homology to the Drosophila Antennapedia probe was restricted to 225 bp between the Xhol sites at positions 39 and 264 in Fig. 1 (hereafter referred to as the Xhol fragment).

A Southern blot of Manduca genomic DNA probed under high-stringency conditions with the Manduca fragment containing the homeobox homology revealed one hybridizing band (Fig. 2), indicating that this clone represents a single copy gene in the Manduca genome. A rescreening of three genomic equivalents of the Manduca library with this same probe yielded an additional 14 homeobox-containing clones, which separated into 9 classes by restriction enzyme mapping. Preliminary sequence analysis of three of these classes indicate that they have significant sequence similarity to the Drosophila genes Deformed, Antennapedia and Ultrabithorax (Booker, in preparation).

Sequence analysis shows that the 225 bp Xhol fragment contains part of a homeobox that is 78 % similar at the nucleotide level and 92 % similar at the amino acid level to the Drosophila Antennapedia homeobox (McGinnis et al. 1984; Schneuwly et al. 1986; Stroeher et al. 1986), which was used as the initial hybridization probe. More importantly, the amino acid sequence of the Manduca homeobox (Fig. 3) is identical to that of the Drosophila abdominal-A homeobox (abd-A) (Karch et al. 1990), although the nucleotide similarity is only 78 %.

The sequence identity goes beyond the homeobox as well. Upstream of the Manduca homeobox the identity extends an additional five amino acids, then ends at an intron-exon splice junction in the Drosophila abd-A gene (see Fig. 3). We have sequenced an additional 1.2 kb upstream and found AT-rich sequences with a low coding sequence probability (data not shown).

At the other end of the homeobox, the similarity of the predicted Manduca ammo acid sequence extends another 21 amino acids beyond the end of the Drosophila homeobox-containing exon (18/21 amino acids are identical; Fig. 3). There is no consensus donor splice site that would correspond to the intron splice junction at the 3 ′ end of the Drosophila abd-A homeobox exon. The Drosophila abd-A gene encodes a glutamine repeat thirty amino acids downstream from the homeobox (27 of the next 32 amino acids are glutamine; Karch et al. 1990). This extended glutamine repeat does not appear in the Manduca sequence, although there is a region 21 amino acids downstream from the homeobox where 5 out of 7 predicted amino acids are glutamine. The sequence similarity between the Manduca and Drosophila genes does not resume within the next 128bp of sequence (data not shown). The sequence similarities both within the homeobox, and extending in either direction, strongly suggest that we have isolated a Manduca homologue of the Drosophila abd-A gene.

To determine the developmental profile and the tissue specificity of the Manduca abd-A gene, we monitored the expression by both northern blots and in situ hybridizations to tissue sections. When the 216 bp clone was hybridized to a northern blot of either total RNA or poly(A) +, only one 4.7kb band was detected (Fig. 4). Transcripts were not detected in the newly oviposited egg, but were visible in the extended germ band (24% development) and throughout embryogenesis, followed by a decline prior to hatching (88% development) (Fig. 4A). A band of the same size was detected during the third and fifth (final) larval instars (Fig. 4A,B). A northern blot containing total RNA from isolated tissues of 5th instar larvae, either during feeding or at the onset of wandering (the beginning of metamorphosis), showed that the 4.7 kb band was present in the epidermis, ventral nerve cord and muscle (data not shown), but not in the Malpighian tubules or fat body (Fig. 4B).

To aid in understanding the tissue sections of Manduca embryos, we have included a brief description of Manduca development from the cellular blastoderm stage through katatrepsis (approximately 50 % of development). A more detailed description of these events can be found in Dorn et al. (1987) and Dow et al. (1988). Events in Manduca embryogenesis will be referred to as the percentage of the developmental time to hatching; at 26 °C embryogenesis takes approximately 92 h.

The early Manduca germ anlage is saddle-shaped and is two times wider (dorsal-ventral axis) than it is long (anterior-posterior axis). Gastrulation involves extensive shape change of the germ anlage. The ventral furrow invaginates first in the presumptive gnathal-thoracic region of the anlage, and proceeds over the length of the embryo in either direction (Fig. 5A,B). As the ventral furrow lengthens, the width of the entire germ anlage is greatly reduced. The germ band extends along the circumference of the egg, so that the head and tail nearly meet on the dorsal side of the egg (20% development), then retracts over the next 20% of development so that the head and tail are now at opposite poles of the egg. The movements of katatrepsis push the embryo across the interior of the egg, changing the flexure of the embryo from convex to concave in relation to the anterior-posterior axis (Fig. 6A: prekatatrepsis, Fig. 7A: postkatatrepsis).

The earliest developmental stage analyzed by in situ hybridization was 12% development (Fig. 5). By this stage, gastrulation has progressed through the anterior abdominal segments while the posterior abdominal segments remain expanded, as seen in the whole mount of the embryo in Fig. 5A,B. In situ hybridizations to frontal sections, using the anti-sense strand of the 216 bp fragment, showed abundant silver grains in the posterior 2/3 of the germ band (Fig. 5C,D,E,F). Signal is visible over both the presumptive ectoderm and mesoderm, but less abundant over the presumptive mesoderm (Fig. 5C,D). The exact limits of the silver grains with respect to segment number cannot be defined, but reconstruction of the serial sections confirms that signal is absent in the anterior portion of the gastrula, as well as in the most extreme posterior regions.

By 40% development, the embryo has completed germ-band elongation and retraction (Fig. 6A). Mouth parts and thoracic limb buds are substantially developed, and the posterior hindgut has extended one-third of the length of the embryo The nervous system is well developed, and the eighth, ninth and tenth abdominal ganglia are beginning to fuse. Hybridization signal from the Manduca abd-A transcripts formed an abrupt border on the midline of the first abdominal segment (Fig. 6C,D). No signal was detectable in the anterior portion of the first abdominal segment or any other more anterior segments (Fig. 6C,D), but was abundant over all the abdominal ganglia, as well as the abdominal ectodermal and mesodermal tissues through the tenth abdominal segment (Fig. 6E,F).

Tissue sections from 40% development (Fig. 6B), as well as from 12% and 50% development (data not shown), were all hybridized to the sense strand of the 216bp fragment. In no case was signal above background detected.

By 50% development, the embryo has completed katatrepsis and begun dorsal closure (Fig. 7A). Note that the posterior-most ganglia in this stage embryo represents a fusion of eighth, ninth and tenth abdominal ganglia. The anterior boundary of expression remains the same at this stage. Transcripts were most abundant over all the abdominal ganglia posterior to the first abdominal segment, and present over the ectoderm and mesoderm of these same segments as well. Silver grains decreased in abundance in the ectoderm posterior to the last ganglia (Fig. 7B,C), and were absent over the most posterior ectodermal tissues (not visible in section shown).

Figs 7 D and E show a transverse section of the same stage embryo. Silver grains were abundant over the abdominal ganglia, epidermis and mesoderm, but were absent over the three paired Malpighian tubules and the hindgut. There were no grains visible above background in the head segments. This tissue specificity in the embryo matches that shown by the northern blot of RNA isolated from tissues from fifth instar and wandering stage larvae (Fig. 4B).

We have isolated and partially sequenced a Manduca homeobox-containing gene. The predicted amino acid sequence of an 86 bp region of this gene is nearly identical to that of the homeobox-containing exon of the Drosophila abd-A gene (Karch et al. 1990). The sequence identity includes the homeobox, an additional five amino acids 5′ to the homeobox, and 18 out of 21 amino acids 3′ to the homeobox. The temporal expression of transcripts from the Manduca gene during both embryogenesis and postembryonic development mimics that from the Drosophila gene. The tissuespecific distribution of transcripts in the germ-band retracted embryo reflects the difference in visible segment number between the Manduca and Drosophila abdomen.

Genes that show a similar degree of identity to the Drosophila abdominal-A gene have now been isolated from four other insect species. Drosophila melanogaster (Karch et al. 1990), Apis mellifera (Walldorf et al. 1989), Manduca sexta (this paper), Schistocerca gregaria (Akam et al. 1988; Tear et al. 1990) and Triboltum sp. (Brown et al. 1990) all share a 72 amino acid stretch that extends in both directions beyond the homeobox. The conservation of this homeobox-containing domain in all 5 species (with the exception of position 22 in Apis’) is remarkable. Treisman et al. (1989) and Hanes and Brent (1989) have shown that the asparagine in position 51 is critical for the DNA-binding specificity of the Drosophila homeobox domain. Yet it is clear that other regions of the domain are also important in forming the protein-DNA complex (Otting et al. 1990; Percival Smith et al. 1990) and in determining its target specificity (Hayashi and Scott, 1990). The resistance of the whole region to change-over evolution underscores its important functional role.

In addition to the amino acid conservation, the intron-exon splice junction 5 ′ to the homeobox also appears to be conserved among Drosophila, Manduca, Schistocerca and Tribohum. The Manduca gene appears to share this exon boundary by virtue of the conserved acceptor splice junction sequence (pyr,pyr, pyr,NAG) (Breathnach and Chambón, 1981) present at this site, and by the fact that the sequences immediately preceding this junction in the Manduca sequence are extremely AT-rich and the coding sequence probability decreases significantly. The Apis abd-A homologue does not appear to share this splice junction, as the sequence homology of the genomic clone continues an additional 17 amino acids 5′ to this junction without any disruption (Walldorf et al. 1989). The splice junction 3′ to the homeobox in Drosophila is not conserved in any of the other insects analyzed.

The sequence homology between the Drosophila abd-A cDNA and the Manduca genomic clones ends at a region that encodes a small glutamine repeat (5 amino acids) in the Manduca sequence. This position is 8 amino acids upstream of a glutamine repeat in the translated Drosophila cDNA. Glutamine-rich regions are present in a number of known transcription factors and in most of the Drosophila homeobox-containing genes although their position in relation to the homeobox varies (Regulski et al. 1985; Laughon et al. 1985). There is evidence from mammalian transcriptional regulators that the presence of a glutamine-rich segment enhances transcriptional activation (Courey and Tjian, 1988), although the significance of the size or position of this segment is as yet unknown.

Preliminary reports indicate that multiple transcripts are produced from the Drosophila abd-A gene: 5.4 and 4.8 kb were reported by Rowe and Akam (1988), and 5.4 and 5.1 kb by Karch et al. (1990). Using a Manduca abd-A homeobox-containing probe, we detect only one 4.7 kb band on a northern blot containing either total or poly(A)+ RNA. This is consistent with the lack of conservation of splice junctions between the Manduca and Drosophila sequences. Further analysis of differential splicing in both the Drosophila and Manduca genes might reveal regulatory changes that have occurred in evolution.

In Drosophila, homeotic genes are thought to have a role in maintenance of segment character throughout the life cycle of the fly (Garcia-Bellido and Lewis, 1976; Morata and Garcia-Bellido, 1976). The expression pattern of the Manduca abd-A gene indicates a similar requirement throughout the life cycle of Manduca’, abdA transcripts first appear in the maturing cellular blastoderm or early gastrula and expression is maintained throughout embryogenesis and larval life in both insects (Karch et al. 1990). The Manduca abd-A transcripts are also distributed amongst similar tissue types as the Drosophila abd-A protein (Karch et al. 1990): they are found in epidermal, mesodermal and neuronal tissues, but not in the Malpighian tubules.

The spatial distribution of transcripts in the late embryo reveals a striking similarity between Manduca and Drosophila. In both insects, the transcripts form an abrupt boundary in the middle of the first abdominal segment, the boundary that marks the beginning of parasegment (ps) 7 in Drosophila. The conservation of this domain of expression for a homeotic gene indicates the conservation of parasegments as domains of gene activity in Manduca. Whether the initial morphological metameric divisions in the Manduca embryo correspond to segmental or parasegmental iterations is not known.

In contrast, the posterior boundary of abdominal-A expression does not occur in the same segment in Manduca and Drosophila. The Drosophila abd-A protein is present from ps7 through the middle of psl3, indicating that its posterior boundary is not parasegmental (Karch et al. 1990; Macias et al. 1990). The posterior boundary to the expression of the transcript is less clearly resolved. In situ hybridization to germband-retracted embryos showed an abundant signal from ps7 to psl2 and a weaker signal in psl3 (Rowe and Akam, 1988). Harding et al. (1985) and Regulski et al. (1985) found an additional weak signal in psl4. In the Manduca embryo, transcripts appear equally abundant over ten abdominal segments (equivalent to ps7-15.5 in Drosophila’), then are absent in the terminalia. Thus, as in Drosophila, the posterior boundary is not parasegmental, but it extends two segments further than in Drosophila. In the Schistocerca embryo, the abd-A protein also initially extends through the tenth abdominal segment, then later in development retracts to the ninth segment (Tear et al. 1990). The observed abd-A expression pattern correlates with the difference in the number of visible segments in the abdomen of these insects: Manduca and Schistocerca embryos have ten well-developed abdominal segments compared to the eight in Drosophila.

The conservation of parasegment 7 as the anterior boundary of abd-A expression and the divergence of the posterior boundary among these three insects suggests that these boundaries may be differently regulated in development and evolution. Although the specific regulation of the onset of transcription of abd-A is not well-characterized in Drosophila, Abd-B mutations affect the limits of abd-A expression so that late in development, abd-A appears in psl4 and 15 (Macias et al. 1990). Thus, Abd-B might be responsible for the late retraction of abd-A in Schistocerca. Other factors apparently act earlier to delimit the initial domain of abd-A expression in Drosophila (Macias et al. 1990) The variation in this domain among these three insects suggests that all of these factors may not be evolution-anly conserved.

Lepidopterans are not easily classified into any of the categories of long, intermediate or short germ insects (see Sander, 1976). Segments appear with an obvious anterior-posterior progression, but it is not known when segmental fates are established: are segments specified in the cellular blastoderm, or are they progressively determined as segments appear during gastrulation? Luscher (1944) was able to cause disruption of individual segments by UV-irradiating transverse strips of the lepidopteran Tineola egg prior to the blastoderm stage. Similarly, Myohara and Kiguchi (1990) were able to construct a segment-specific fate map for Bombyx mori within 2 h of egg deposition (the cellular blastoderm is normally formed about 12 h). Although these results do not necessarily mean that determination has already occurred at the time of irradiation (see Sander, 1976 for discussion), they do show that the anlage for each abdominal segment can be mapped at the cellular blastoderm stage and does not undergo subsequent expansion Carr and Taghert (1989), using the onset of Manduca fasciculin II (TNI) as a marker, suggest that segments are progressively determined as gastrulation proceeds in Manduca. Our experiments show an abundance of transcripts in the presumptive abdominal region at 12 % of development, before this region has undergone the movements of gastrulation. This result suggests that, at least in relation to abd-A expression, the abdominal segments have some information as to their character before they are visibly distinguishable as abdominal segments.

We have shown that there is an extremely high degree of conservation of the amino acid sequences of the homeotic gene abdominal-A. Not only is the sequence conserved, but two of the principles of homeotic gene function -a parasegmental domain of gene expression and persistence of expression throughout the life of the insect -also appear to be conserved. The differences that we have detected appear in the regulation of the abdominal-A gene: its initial embryonic distribution is over a greater number of segments and, secondly, differential splicing of the mRNA is not detectable. It is not surprising to find the homeotic genes conserved in other insects but, nonetheless, the generation of morphologically diverse insects that respond differently to experimental perturbations remains a mystery. Our results suggest that the regulation of the homeotic gene products may begin to account for these differences.

We thank Frank Horodyski and John Rebers for their undying patience and expert technical advice; M. Akam for pointing out the abd-A homology from unpublished sequences; R. Garber for the Drosophila probes, and G. Schubiger and B. Wakimoto for their comments on a draft of this manuscript. This study was supported by NIH AI12459 RB was supported by NIH NS-13079. LMN was supported by NIH Training Grant GM072780